CASSIA: a multi-agent large language model for reference free, interpretable, and automated cell annotation of single-cell RNA-sequencing data
Open accessAbstract
Cell type annotation is an essential step in single-cell RNA-sequencing analysis, and numerous annotation methods are available. Most require a combination of computational and domain-specific expertise, and they frequently yield inconsistent results that can be challenging to interpret. Large language models have the potential to expand accessibility while reducing manual input and improving accuracy, but existing approaches suffer from hyperconfidence, hallucinations, and lack of reasoning. To address these limitations, we developed CASSIA for automated, accurate, and interpretable cell annotation of single-cell RNA-sequencing data. As demonstrated in analyses of over 970 cell types, CASSIA improves annotation accuracy in benchmark datasets as well as complex and rare cell populations, and also provides users with reasoning and quality assessment to ensure interpretability, guard against hallucinations, and calibrate confidence.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Multi-agent systems were predominantly role-based, with agents assigned to cognitively separable subtasks such as drafting, verification, retrieval, synthesis, supervision, or reporting.
“Multi-agent systems were predominantly role-based, with agents assigned to cognitively separable subtasks such as drafting, verification, retrieval, synthesis, supervision, or reporting [22,26,27].”
From Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review (2026) · cites this paper as [26]
The Online Methods and Figure 1 description assign CASSIA’s agents distinct annotation, validation, formatting, scoring, and reporting tasks; the RAG pipeline adds retrieval-related tasks. This supports CASSIA as an example of role-based task separation, but the paper does not establish that multi-agent systems were predominantly organized this way. (LLM, conf=0.85)
Experimental — accuracy not yet measuredMost systems incorporated example-based grounding, operationalized as explicit links to retrieved clinical guidelines, biomedical literature, databases, ontologies, or knowledge graph paths.
“In addition, most systems incorporated example-based grounding, operationalized as explicit links to retrieved clinical guidelines, biomedical literature, databases, ontologies, or knowledge graph path [19,25,26].”
From Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review (2026) · cites this paper as [26]
The Retrieval-Augmented Generation (RAG) pipeline section says CASSIA can optionally query a cell-marker database and use biological ontologies to inform annotations. It does not establish that most systems use this approach or that CASSIA provides explicit links to retrieved sources; the other listed source types are not described as part of its pipeline. (LLM, conf=0.82)
Experimental — accuracy not yet measuredSystems focused on CDSSs, screening, and biomedical reasoning primarily relied on a reasoning-plus-evidence paradigm, combining textual rationales with explicit grounding in retrieved sources.
“Systems focused on CDSSs, screening, and biomedical reasoning (e.g., ENTAgents, DrugAgent, DSM5AgentFlow, BRAD, and CASSIA) primarily relied on a “reasoning + evidence” paradigm, combining textual rationales with explicit grounding in retrieved sources [19,24,25,26,27].”
From Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review (2026) · cites this paper as [26]
The CASSIA implementation and Reporter agent sections describe chain-of-thought annotation and a report containing reasoning and supporting evidence. But the Retrieval-Augmented Generation pipeline section says database retrieval is optional, not something CASSIA primarily relies on; the paper also concerns single-cell annotation rather than CDSSs or screening. No page numbers are available in our copy. (LLM, conf=0.86)
Experimental — accuracy not yet measuredNine studies met all inclusion criteria and were included in the final synthesis.
“Nine studies met the inclusion criteria.”
From Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review (2026) · cites this paper as [26]
Other included studies incorporated retrospective validation using benchmark or real-world datasets (ADAM-1, DrugAgent, EMAI, and CASSIA).
“In contrast, others incorporated retrospective validation using benchmark or real-world datasets (e.g., ADAM-1, DrugAgent, EMAI, and CASSIA) [19,22,23,26].”
From Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review (2026) · cites this paper as [26]
Retrieval-augmented generation was frequently incorporated to ground reasoning in external evidence, particularly in CBM-RAG, DrugAgent, DSM5AgentFlow, CASSIA, and BRAD.
“Retrieval-augmented generation (RAG) was frequently incorporated to ground reasoning in external evidence, particularly in systems aiming to mitigate hallucinations or enhance reliability, such as CBM-RAG, DrugAgent, DSM5AgentFlow, CASSIA, and BRAD [19,20,24,25,26].”
From Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review (2026) · cites this paper as [26]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Aran, D., Looney, A. P., Liu, L., Wu, E., Fong, V., Hsu, A., Chak, S., Naikawadi, R. P., Wolters, P. J., Abate, A. R., et al. (2019) · Nature Immunology · doi:10.1038/s41590-018-0276-y · Cited by 4,984 · see citing works
Huang, L., Yu, W., Ma, W., Zhong, W., Feng, Z., Wang, H., Chen, Q., Peng, W., Feng, X., Qin, B., et al. (2023) · arXiv · doi:10.48550/arxiv.2311.05232 · Cited by 2,516 · see citing works
Huang, L., Yu, W., Ma, W., Zhong, W., Feng, Z., Wang, H., Chen, Q., Peng, W., Feng, X., Qin, B., et al. (2025) · ACM Transactions on Information Systems · doi:10.1145/3703155 · Cited by 2,142 · see citing works
Large Language Models are Zero-Shot Reasoners
Kojima, T., Gu, S. S., Reid, M., Matsuo, Y., & Iwasawa, Y. (2022) · arXiv · doi:10.48550/arxiv.2205.11916 · Cited by 2,013 · see citing works
Large language models are zero-shot reasoners
Kojima, T., Gu, S. (., Reid, M., Matsuo, Y., & Iwasawa, Y. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1613 · Cited by 2,013 · see citing works
CellMarker: a manually curated resource of cell markers in human and mouse
Zhang, X., Lan, Y., Xu, J., Quan, F., Zhao, E., Deng, C., Luo, T., Xu, L., Liao, G., Yan, M., et al. (2019) · Nucleic Acids Research · doi:10.1093/nar/gky900 · Cited by 1,524 · see citing works
scGPT: Towards Building a Foundation Model for Single-Cell Multi-omics Using Generative AI
Cui, H., Wang, C., Maan, H., Pang, K., Luo, F., & Wang, B. (2023) · doi:10.1101/2023.04.30.538439 · Cited by 1,383 · see citing works
Cross-tissue immune cell analysis reveals tissue-specific features in humans
Conde, C. D., Xu, C., Jarvis, L. B., Rainbow, D. B., Wells, S. B., Gomes, T., Howlett, S. K., Suchánek, O., Polański, K., King, H. W., et al. (2022) · Science · doi:10.1126/science.abl5197 · Cited by 1,165 · see citing works
Detecting hallucinations in large language models using semantic entropy
Farquhar, S., Kossen, J., Kuhn, L., & Gal, Y. (2024) · Nature · doi:10.1038/s41586-024-07421-0 · Cited by 934 · see citing works
Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language ModelsPreprint
Wang, L., Xu, W., Lan, Y.-H., Hu, Z.-Q., Lan, Y.-S., Lee, R. K.-W., & Lim, E.-P. (2023) · Annual Meeting of the Association for Computational Linguistics · doi:10.48550/arxiv.2305.04091 · Cited by 917 · see citing works
Ianevski, A., Giri, A. K., & Aittokallio, T. A. (2022) · Nature Communications · doi:10.1038/s41467-022-28803-w · Cited by 763 · see citing works
scmap: projection of single-cell RNA-seq data across data sets
(2018) · doi:10.1038/nmeth.4644 · Cited by 735 · see citing works
A comparison of automatic cell identification methods for single-cell RNA sequencing data
Abdelaal, T., Michielsen, L., Cats, D., Hoogduin, D., Mei, H., Reinders, M. J. T., & Mahfouz, A. (2019) · Genome Biology · doi:10.1186/s13059-019-1795-z · Cited by 628 · see citing works
Yang, F., Wang, W., Wang, F., Fang, Y., Tang, D., Huang, J., Lu, H., & Yao, J. (2021) · doi:10.1101/2021.12.05.471261 · Cited by 617 · see citing works
The TRIPOD-LLM reporting guideline for studies using large language models
Gallifant, J., Afshar, M., Ameen, S., Aphinyanaphongs, Y., Chen, S., Cacciamani, G., Demner-Fushman, D., Dligach, D., Daneshjou, R., Fernandes, C., et al. (2025) · Nature Medicine · doi:10.1038/s41591-024-03425-5 · Cited by 615 · see citing works
Role play with large language models
Shanahan, M., McDonell, K., & Reynolds, L. (2023) · Nature · doi:10.1038/s41586-023-06647-8 · Cited by 491 · see citing works
scPred: accurate supervised method for cell-type classification from single-cell RNA-seq data
Alquicira-Hernández, J., Sathe, A., Ji, H. P., Nguyen, Q. H., & Powell, J. E. (2019) · Genome biology · doi:10.1186/s13059-019-1862-5 · Cited by 474 · see citing works
Lamy, J.-B. (2017) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2017.07.002 · Cited by 366 · see citing works
Tan, Y., & Cahan, P. (2019) · Cell Systems · doi:10.1016/j.cels.2019.06.004 · Cited by 352 · see citing works
The Cell Ontology 2016: enhanced content, modularization, and ontology interoperability
Diehl, A. D., Meehan, T. F., Bradford, Y. M., Brush, M., Dahdul, W., Dougall, D. S., He, Y. O., Osumi-Sutherland, D., Ruttenberg, A., Sarntivijai, S., et al. (2016) · Journal of Biomedical Semantics · doi:10.1186/s13326-016-0088-7 · Cited by 332 · see citing works
scCATCH: Automatic Annotation on Cell Types of Clusters from Single-Cell RNA Sequencing Data
Shao, X., Liao, J., Lu, X., Xue, R., Ai, N., & Fan, X. (2020) · iScience · doi:10.1016/j.isci.2020.100882 · Cited by 289 · see citing works
SCINA: A Semi-Supervised Subtyping Algorithm of Single Cells and Bulk Samples
Zhang, Z., Luo, D., Zhong, X. Y., Choi, J. H., Ma, Y., Wang, S., Mahrt, E., Guo, W., Stawiski, E. W., Modrusan, Z. D., et al. (2019) · Genes · doi:10.3390/genes10070531 · Cited by 281 · see citing works
Assessing GPT-4 for cell type annotation in single-cell RNA-seq analysis
Hou, W., & Ji, Z. (2024) · Nature Methods · doi:10.1038/s41592-024-02235-4 · Cited by 277 · see citing works
CHETAH: a selective, hierarchical cell type identification method for single-cell RNA sequencing
de Kanter, J. K., Lijnzaad, P., Candelli, T., Margaritis, T., & Holstege, F. C. P. (2019) · Nucleic Acids Research · doi:10.1093/nar/gkz543 · Cited by 274 · see citing works
Automated methods for cell type annotation on scRNA-seq data
Pasquini, G., Rojo Arias, J. E., Schäfer, P., & Busskamp, V. (2021) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2021.01.015 · Cited by 249 · see citing works
A Review of Single-Cell RNA-Seq Annotation, Integration, and Cell–Cell Communication
Cheng, C., Chen, W., Jin, H., & Chen, X. '. (2023) · Cells · doi:10.3390/cells12151970 · Cited by 237 · see citing works
Clarke, Z. A., Andrews, T. S., Atif, J., Pouyabahar, D., Innes, B. T., MacParland, S. A., & Bader, G. D. (2021) · Nature Protocols · doi:10.1038/s41596-021-00534-0 · Cited by 219 · see citing works
Large Language Models are Better Reasoners with Self-Verification
Weng, Y., Zhu, M., Xia, F., Li, B., He, S., Liu, S., Sun, B., Liu, K., & Zhao, J. (2023) · Findings of the Association for Computational Linguistics: EMNLP 2023 · doi:10.18653/v1/2023.findings-emnlp.167 · Cited by 110 · see citing works
Large Language Models are Better Reasoners with Self-Verification
Weng, Y., Zhu, M., Xia, F., Li, B., He, S., Liu, S., Sun, B., Liu, K., & Zhao, J. (2022) · arXiv · doi:10.48550/arxiv.2212.09561 · Cited by 110 · see citing works
Harnessing large language models (LLMs) for candidate gene prioritization and selection
Toufiq, M., Rinchai, D., Bettacchioli, É., Kabeer, B. S. A., Khan, T., Subba, B., White, O., Yurieva, M., George, J., Jourde‐Chiche, N., et al. (2023) · Journal of Translational Medicine · doi:10.1186/s12967-023-04576-8 · Cited by 64 · see citing works
Kim, J., Wang, K., Weng, C., & Liu, C. (2024) · The American Journal of Human Genetics · doi:10.1016/j.ajhg.2024.08.010 · Cited by 53 · see citing works
Principled Instructions Are All You Need for Questioning LLaMA-1/2, GPT-3.5/4Preprint
Bsharat, S. M., Myrzakhan, A., & Shen, Z. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2312.16171 · Cited by 51 · see citing works
The long but necessary road to responsible use of large language models in healthcare research
Kwong, J. C. C., Wang, S., Nickel, G., Cacciamani, G. E., & Kvedar, J. C. (2024) · npj Digital Medicine · doi:10.1038/s41746-024-01180-y · Cited by 32 · see citing works
ChatGPT is a black box: how AI research can break it open
(2023) · Nature · doi:10.1038/d41586-023-02366-2 · Cited by 29 · see citing works
Mixture-of-Agents Enhances Large Language Model Capabilities
Wang, J., Wang, J., Athiwaratkun, B., Zhang, C., & Zou, J. (2024) · arXiv · doi:10.48550/arxiv.2406.04692 · Cited by 25 · see citing works
Li, J., Zhang, Q., Yu, Y., Fu, Q., & Ye, D. (2024) · arXiv · doi:10.48550/arxiv.2402.05120 · Cited by 22 · see citing works
Fu, Q., Dong, C., Liu, Y., Xia, X., Liu, G., Zhong, F., & Liu, L. (2024) · Briefings in Bioinformatics · doi:10.1093/bib/bbae392 · Cited by 18 · see citing works
Gill, J. K., Chetty, M., Lim, S., & Hallinan, J. S. (2024) · PLoS ONE · doi:10.1371/journal.pone.0303231 · Cited by 14 · see citing works
Shireman, J. M., White, Q., Ni, Z., Mohanty, C., Cai, Y., Zhao, L., Agrawal, N., Gonugunta, N., Wang, X., Mccarthy, L., et al. (2024) · iScience · doi:10.1016/j.isci.2024.109601 · Cited by 7 · see citing works
Self-citation
Flagged first, then by citations
Research works printed but not linked: 2
- which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
- which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Cited by
Sorted by citations
Extracellular Matrix Deposition Controls Early Differentiation Patterns in Human Cardiac GastruloidsPreprint
Marchand, M., Sahni, G., Arora, A., Serhrouchni, S., Dilasser, F., Monet, A., Lucciani, F., Rajendiran, H., Chabot, L., Raffi, S. B. M., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.03.18.643854 · Cited by 1 · see citing works
Explainable Agentic Artificial Intelligence in Healthcare: A Scoping Review
Collaco, B. G., Prabha, S., Gomez-Cabello, C. A., Haider, S. A., Genovese, A., Wood, N. G., Gopala, N., Raman, R., Hester, E. O., & Forte, A. J. (2026) · Bioengineering · doi:10.3390/bioengineering13050513 · Cited by 3 · see citing works
“For example, CBM-RAG separated disease-specific reasoning from report synthesis [20]; ADAM-1 distinguished computational analysis from narrative interpretation [22]; and CASSIA partitioned annotation, validation, quality scoring, and reporting into discrete stages [26].”
InfluentialSurveying Multi-agent Genomic Systems with a Spotlight on Oncology
Mannai, Z., Omheni, N., & Mahmoudi, R. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-32023-0_51
- Authors
- Xie, E., Cheng, L., Shireman, J., Cai, Y., Liu, J., Mohanty, C., Dey, M., & Kendziorski, C.
- Affiliations
- University of Wisconsin–Madison (US)
- Corresponding author
- Kendziorski, Christina
- ORCID
- Venue
- bioRxiv (Cold Spring Harbor Laboratory)
- Publisher
- openRxiv
- Published
- 2024-12-08
- OpenAlex
- W4405202975
- Cited by
- 3
- References
- 35
- Citations by year
- 2026: 2 · 2025: 1
- Type
- posted content
- Language
- en
- Open access
- green
- doi.org
- Locations
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion)
- Funders
- National Institutes of Health
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Cell Image Analysis Techniques
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
- Keywords
- cell type annotation, single-cell RNA sequencing, large language models, multi-agent systems, interpretability, hallucination mitigation, confidence calibration, rare cell populations, automatic annotation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
